本文へ移動
cccskills
無料GitHub で公開

hyperframes-testimonial-builder

Turn real customer reviews, quotes, and case-study results into polished social-proof videos built in HyperFrames -- single-testimonial spotlights, review-wall montages, and stat-driven proof reels with captions and voiceover, cut for vertical, square, and wide. The highest-converting content most businesses never make.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md4.0 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

HyperFrames Testimonial Builder

Nothing sells like a customer saying it. This skill takes the proof a business already has -- Google reviews, testimonial emails, case-study numbers, NPS comments -- and directs it into renderable HyperFrames video. It owns the selection and the story; for composition mechanics, captions, transitions, and TTS, invoke the hyperframes skill (hyperframes-cli for init/lint/preview/render, hyperframes-media for voiceover).

Step 1 -- Gather the proof

Take whatever exists: pasted reviews, a review export, customer-review-aggregator output, case-study docs, or a profile URL to pull from. For each candidate quote capture the text verbatim, the name/attribution available, the specific result mentioned, and the source.

Select for specificity: "They saved us $40K and two weeks" outperforms "Great company, highly recommend" every time. Rank candidates by concrete detail, emotional arc (skeptic-to-believer beats always-happy), and relevance to the audience the video targets.

Step 2 -- Pick the format

  • Spotlight (20-30s) -- one strong story: setup (the problem in the customer's words), the turn, the result, CTA.
  • Review wall (15-25s) -- 4-6 short quotes in rhythm, star ratings and names, momentum building to an aggregate line ("300+ five-star reviews") and CTA.
  • Proof reel (30-45s) -- stats-forward: animated counters for the numbers (customers served, dollars saved, rating), interleaved with the two best quote lines.

Step 3 -- Build the composition

  • Quote text is the hero visual: large, readable, animated in phrase by phrase, timed to the VO. Attribution (first name, business/role, star row) small and consistent.
  • Verbatim always -- trim with ellipses where needed, never rewrite a customer's words into marketing-speak; the texture of real speech is the credibility.
  • Brand rules honored: the business's palette and type (reuse a claude-design-system-architect or design-style-installer token set if present); no purple, no emoji.
  • Captions synced (muted autoplay is the default viewing mode), counters and star fills animated deterministically per HyperFrames seek-safety rules, VO via the media pipeline reading the quotes with a neutral warm voice -- or silence with music-bed timing if the user prefers.
  • Lint and preview before declaring done; never claim it renders without running the pipeline.

Step 4 -- Deliver

Platform cuts (9:16, 1:1, 16:9), a thumbnail frame, and the caption text for the post. Note where each cut runs best: review wall for the Google Business Profile and Instagram, spotlight for the website hero and proposals, proof reel for paid.

Rules

  • Real quotes from real customers only. Never fabricate, composite, or "improve" a testimonial -- fabricated social proof is a legal problem, not a style choice.
  • Get-permission reminder in the deliverable: quotes from private emails need the customer's OK before public use; public reviews are fair to feature.
  • Attribution as strong as permitted: full name and business beats first name beats anonymous, in that order of asking.
  • One result number per beat -- stacked stats blur; sequenced stats land.
  • The CTA matches where the proof points: reviews about service -> book a call; reviews about results -> see pricing.

Quick Commands

  • "Testimonial video from [reviews/file]" -- full workflow, format recommended
  • "Review wall from my Google reviews" -- wall format from a profile or export
  • "Spotlight [customer story]" -- single-story treatment
  • "Proof reel from [case study]" -- stats-forward cut

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Audit websites for accessibility issues and WCAG compliance. Use when checking accessibility, fixing a11y issues, or ensuring WCAG compliance.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3362026年10月2日 更新

Deploy a 2-layer parallel agent hierarchy for large, parallelizable work — big refactors, multi-file migrations, codebase-wide audits, bulk generation. A top-tier commander (Fable or Opus) orchestrates the swarms; the user picks a power level (Max Power / Heavy / Balanced / Economy) that sets the Opus/Sonnet/Haiku model mix per layer. Layer 1 is 3-50+ specialist agents, each with its own full context window; Layer 2 is 2+ sub-agents per member. Includes git safety, tiered sizing, a pre-deploy gate, phantom-completion checks, and multi-wave follow-up.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3362026年10月2日 更新

Use to test whether an AI agent, prompt, or skill actually works before or after it ships — build a small eval set, score runs, and catch regressions when the prompt or model changes. Trigger when the user asks whether a prompt is good, wants to compare models or versions, sees inconsistent agent output, or is about to put an agent in front of customers.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3362026年10月2日 更新

Deploys swarms of sub-agents for massive parallel data processing tasks. Unlike agent-army (which is for code changes), this is for DATA tasks -- processing 1000 documents, analyzing datasets, bulk content generation. Configurable swarm size, task distribution, result aggregation, progress tracking, and error recovery.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3362026年10月2日 更新

Designs and deploys custom agent teams for specific business workflows. Interactive discovery of business processes, then generates complete team configurations with specialized agent roles, tool access, communication protocols, and handoff rules.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3362026年10月2日 更新

Agent-to-Agent (A2A) communication protocol. Connect two or more Claude agents that pass messages, share context, delegate tasks, and collaborate. Implements structured handoffs, shared memory, and multi-agent conversations.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3362026年10月2日 更新

OneWave-AI のスキルをすべて見る

このスキルの問題を報告する